Naspolini et al. 2024 — Heavy metals in human milk, São Paulo cohort
Naspolini et al. measured As, Pb, Hg, and Cd in 185 human milk samples from the Germina prospective cohort in São Paulo, Brazil, collected at 3 months postpartum, and assessed associations with infant neurodevelopment via the Bayley-III scales at 3, 5–9, and 10–16 months of age. Arsenic had the highest detection rate (38.6%, mean 2.76 µg/L), followed by Hg (23.9%, mean 1.96 µg/L) and Pb (22.8%, mean 2.09 µg/L); Cd was not detected above LOQ in any sample. Infants exposed to Pb in human milk showed significantly lower language development trajectories at 10–16 months compared to non-exposed infants (β = −0.413; 95% CI −0.653 to −0.173), even though mean Pb concentrations in this cohort were lower than in most prior Brazilian and international studies.
Key numbers
Arsenic (tAs): n=185; detection rate 38.6% (71/185); mean (SD) 2.76 (4.09) µg/L; range 0.10–34.55 µg/L. LOQ 0.06 µg/L. Note: speciation not performed; these are total As by ICP-MS.
Mercury (tHg): n=185; detection rate 23.9% (44/185); mean (SD) 1.96 (6.68) µg/L; range 0.04–54.41 µg/L. LOQ 0.3 µg/L. Note: speciation not performed; total Hg by ICP-MS.
Lead (Pb): n=185; detection rate 22.8% (42/185); mean (SD) 2.09 (5.36) µg/L; range 0.15–40.90 µg/L. LOQ 1 µg/L. Among 42 exposed infants, mean Pb 8.67 (8.43) µg/L. At 10–16 months, exposed infants had mean language Bayley score 97.47 (13.36) vs. non-exposed 102.96 (11.53) (p<0.001).
Cadmium (Cd): n=185; 0 detections (all below LOQ of 0.08 µg/L).
Neurodevelopmental outcome: Pb exposure × infant age interaction: β = −0.413 (95% CI −0.653 to −0.173, p<0.001) in a linear mixed-effects model adjusted for infant age, socioeconomic status, maternal education, infant sex, and pseudoweights for population representativeness.
Methods
Germina cohort, São Paulo metropolitan area, 2021–2022. Analysis by ICP-MS (PerkinElmer NexION 300D) at INCOS/Fiocruz, Rio de Janeiro. Acid digestion with 65% nitric acid. LOD/LOQ: Pb 0.015/1 µg/L; Hg 0.007/0.3 µg/L; Cd 0.002/0.08 µg/L; As 0.003/0.06 µg/L. Values below LOQ imputed as LOD/2. Since imputation rate for Pb exceeded 50%, infants were classified dichotomously as exposed vs. non-exposed for statistical analysis. Neurodevelopment assessed by Bayley-III at three time points (3 months; 5–9 months; 10–16 months). Longitudinal analysis used linear mixed-effects model (nlme R package). Basis: liquid (µg/L in whole human milk).
Implications
Certification: Human milk is not a certified product category, but Pb levels in human milk reflect maternal body burden from dietary and environmental exposures. These data anchor the exposure side of the infant-food Pb risk picture: infants exposed to as little as mean 8.67 µg/L Pb in human milk show measurable language developmental deficits, setting context for why infant food Pb limits matter.
Health: Supports no-safe-level Pb exposure framework for infants; language effects observed at concentrations lower than in most prior Brazilian state studies, consistent with the CDC/WHO consensus that no threshold exists for Pb neurodevelopmental toxicity.
Courses: Demonstrates that Pb, tAs, and tHg contamination of human milk occurs in urban Brazil; Cd was below detection, providing a contrast to other matrices. Germina cohort design provides longitudinal neurodevelopmental outcome data paired with exposure measurement.
App/Microbiome: Out of direct product scope but relevant to exposure modeling for breastfed infants.